In this article, we’re going show
you how easy it is to move from connecting to the database holding your
data to producing the results you need. It’s meant to be a quick and
friendly introduction to {dm}, so it is low on details and caveats.
Links to detailed documentation are provided at the end. (If your data
is in data frames instead of a database and you’re in a hurry, jump over
to vignette("howto-dm-df")
.)
dm objects can be created from individual tables or loaded directly from a relational data model on an RDBMS (relational database management system).
For this demonstration, we’re going to work with a model hosted on a public server. The first thing we need is a connection to the RDBMS hosting the data.
library(RMariaDB)
fin_db <- dbConnect(
MariaDB(),
username = "guest",
password = "ctu-relational",
dbname = "Financial_ijs",
host = "relational.fel.cvut.cz"
)
We create a dm object from an RDBMS using dm_from_con()
,
passing in the connection object we just created as the first
argument.
The dm object interrogates the RDBMS for table and column information, and primary and foreign keys. Currently, primary and foreign keys are only available from SQL Server, Postgres and MariaDB.
The dm object can be accessed like a named list of tables:
Additionally, most dm
functions are pipe-friendly and support
tidy
evaluation. We can use [
or the
dm_select_tbl()
verb to derive a smaller dm with the
loans
, accounts
, districts
and
trans
tables:
In many cases, dm_from_con()
already returns a dm with
all keys set. If not, dm allows us to define primary and foreign keys
ourselves. For this, we use learn_keys = FALSE
to obtain a
dm
object with only the tables.
library(dm)
fin_dm_small <-
dm_from_con(fin_db, learn_keys = FALSE) %>%
dm_select_tbl(loans, accounts, districts, trans)
In our data model, id
columns uniquely identify records
in the accounts
and loans
tables, and was used
as a primary key. A primary key is defined with
dm_add_pk()
. Each loan is linked to one account via the
account_id
column in the loans
table, the
relationship is established with dm_add_fk()
.
fin_dm_keys <-
fin_dm_small %>%
dm_add_pk(table = accounts, columns = id) %>%
dm_add_pk(loans, id) %>%
dm_add_fk(table = loans, columns = account_id, ref_table = accounts) %>%
dm_add_pk(trans, id) %>%
dm_add_fk(trans, account_id, accounts) %>%
dm_add_pk(districts, id) %>%
dm_add_fk(accounts, district_id, districts)
Having a diagram of the data model is the quickest way to verify we’re on the right track. We can display a visual summary of the dm at any time. The default is to display the table name, any defined keys, and their links to other tables.
Visualizing the dm in its current state, we can see the keys we have created and how they link the tables together. Color guides the eye.
If we want to perform modeling or analysis on this relational model,
we need to transform it into a tabular format that R functions can work
with. With the argument recursive = TRUE
,
dm_flatten_to_tbl()
will automatically follow foreign keys
across tables to gather all the available columns into a single
table.
Apart from the rows printed above, no data has been fetched from the
database. Use select()
to reduce the number of columns
fetched, and collect()
to retrieve the entire result for
local processing.
We don’t need to take the extra step of exporting the data to work with it. Through the dm object, we have complete access to dplyr’s data manipulation verbs. These operate on the data within individual tables.
To work with a particular table we use dm_zoom_to()
to
set the context to our chosen table. Then we can perform any of the
dplyr operations we want.
fin_dm_total <-
fin_dm_keys %>%
dm_zoom_to(loans) %>%
group_by(account_id) %>%
summarize(total_amount = sum(amount, na.rm = TRUE)) %>%
ungroup() %>%
dm_insert_zoomed("total_loans")
fin_dm_total$total_loans
Note that, in the above example, we use
dm_insert_zoomed()
to add the results as a new table to our
data model. This table is temporary and will be deleted when our session
ends. If you want to make permanent changes to your data model on an
RDBMS, please see the “Persisting results” section in
vignette("howto-dm-db")
.
It’s always smart to check that your data model follows its specifications. When building our own model or changing existing models by adding tables or keys, it is even more important that the new model is validated.
dm_examine_constraints()
checks all primary and foreign
keys and reports if they violate their expected constraints.
For more on constraint checking, including cardinality, finding
candidate columns for keys, and normalization, see
vignette("tech-dm-low-level")
.
Now that you have been introduced to the basic operation of dm, the next step is to learn more about the dm methods that your particular use case requires.
Is your data in an RDBMS? Then move on to
vignette("howto-dm-db")
for a more detailed look at working
with an existing relational data model.
If your data is in data frames, then you want to read
vignette("howto-dm-df")
next.
If you would like to know more about relational data models in order
to get the most out of dm, check out
vignette("howto-dm-theory")
.
If you’re familiar with relational data models, but want to know how
to work with them in dm, then any of
vignette("tech-dm-join")
,
vignette("tech-dm-filter")
, or
vignette("tech-dm-zoom")
is a good next step.
The {dm} package follows the tidyverse principles:
dm
objects are immutable (your data will never be
overwritten in place)dm
objects are pipeable (i.e.,
return new dm
or table objects)The {dm} package builds heavily upon the {datamodelr} package, and upon the tidyverse. We’re looking forward to a good collaboration!
The {polyply} package has a similar intent with a slightly different interface.
The {data.cube}
package has quite the same intent using array
-like
interface.
Articles in the {rquery} package discuss join controllers and join dependency sorting, with the intent to move the declaration of table relationships from code to data.
The {tidygraph}
package stores a network as two related tables of nodes
and edges
, compatible with {dplyr} workflows.
In object-oriented programming languages, object-relational mapping is a similar concept that attempts to map a set of related tables to a class hierarchy.